Abstract
Generative artificial intelligence (GenAI) is increasingly used as an “answer engine” in higher education, raising risks of plausibility replacing verifiability. This study tested whether a semester-long GenAI media literacy module improves undergraduates’ procedural skills of searching, verifying, and interpreting AI-mediated information. Using a quasi-experimental non-equivalent groups pretest–posttest design (T0–T1), we compared two cohorts from L.N. Gumilyov Eurasian National University and Astana International University in Astana, Kazakhstan (N = 135; EG n = 70; CG n = 65). The experimental group completed a 13-week module emphasising primary-source retrieval, lateral reading, evidential reasoning, and responsible attribution; the control group followed the standard curriculum. Outcomes included the 15-item GenAI Media Literacy Questionnaire (GMLQ; 1–7) and a performance task requiring verification of three GenAI-generated claims scored with a rubric (0–27). ANCOVA models adjusted for baseline scores and covariates (age, gender, language of instruction, GPA, baseline GenAI-use frequency) showed a post-test advantage for the experimental group: adjusted EG–CG differences were +0.63 on GMLQ total and +4.67 on performance. Positive effects were consistent across GMLQ subscales and performance components (Search, Verify, Interpret). The findings support embedding sustained, procedure-focused GenAI media literacy instruction in pre-service teacher education to strengthen evidence-based practices.